000888741 001__ 888741
000888741 005__ 20240313103116.0
000888741 0247_ $$2Handle$$a2128/26707
000888741 037__ $$aFZJ-2020-05175
000888741 1001_ $$0P:(DE-Juel1)180150$$aFischer, Kirsten$$b0$$eCorresponding author
000888741 245__ $$aDecomposition of Deep Neural Networks into Correlation Functions$$f2019-10-01 - 2020-10-08
000888741 260__ $$c2020
000888741 300__ $$a91 p.
000888741 3367_ $$2DataCite$$aOutput Types/Supervised Student Publication
000888741 3367_ $$02$$2EndNote$$aThesis
000888741 3367_ $$2BibTeX$$aMASTERSTHESIS
000888741 3367_ $$2DRIVER$$amasterThesis
000888741 3367_ $$0PUB:(DE-HGF)19$$2PUB:(DE-HGF)$$aMaster Thesis$$bmaster$$mmaster$$s1610438050_25886
000888741 3367_ $$2ORCID$$aSUPERVISED_STUDENT_PUBLICATION
000888741 502__ $$aMasterarbeit, RWTH Aachen University, 2020$$bMasterarbeit$$cRWTH Aachen University$$d2020
000888741 520__ $$aRecent years have shown a great success of deep neural networks. One active field of research investigates the functioning mechanisms of such networks with respect to the network expressivity as well as information processing within the network. In this thesis, we describe the input-output mapping implemented by deep neural networks in terms of correlation functions. To trace the transformation of correlation functions within neural networks, we make use of methods from statistical physics. Using a quadratic approximation for non-linear activation functions, we obtain recursive relations in a perturbative manner by means of Feynman diagrams. Our results yield a characterization of the network as a non-linear mapping of mean and covariance, which can be extended by including corrections from higher order correlations. Furthermore, re-expressing the training objective in terms of data correlations allows us to study their role for solutions to a given task. First, we investigate an adaptation of the XOR problem, in which case the solutions implemented by neural networks can largely be described in terms of mean and covariance of each class. Furthermore, we study the MNIST database as an example of a non-synthetic dataset. For MNIST, solutions based on empirical estimates for mean and covariance of each class already capture a large amount of the variability within the dataset, but still exhibit a non-negligible performance gap in comparison to solutions based on the actual dataset. Lastly, we introduce an example task where higher order correlations exclusively encode class membership, which allows us to explore their role for solutions found by neural networks. Finally, our framework also allows us to make predictions regarding the correlation functions that are inferable from data, yielding insights into the network expressivity. This work thereby creates a link between statistical physics and machine learning, aiming towards explainable AI.
000888741 536__ $$0G:(DE-HGF)POF3-574$$a574 - Theory, modelling and simulation (POF3-574)$$cPOF3-574$$fPOF III$$x0
000888741 536__ $$0G:(DE-Juel-1)BMBF-01IS19077A$$aRenormalizedFlows - Transparent Deep Learning with Renormalized Flows (BMBF-01IS19077A)$$cBMBF-01IS19077A$$x1
000888741 536__ $$0G:(DE-Juel1)HGF-SMHB-2014-2018$$aMSNN - Theory of multi-scale neuronal networks (HGF-SMHB-2014-2018)$$cHGF-SMHB-2014-2018$$fMSNN$$x2
000888741 536__ $$0G:(DE-82)EXS-SF-neuroIC002$$aneuroIC002 - Recurrence and stochasticity for neuro-inspired computation (EXS-SF-neuroIC002)$$cEXS-SF-neuroIC002$$x3
000888741 7001_ $$0P:(DE-Juel1)144806$$aHelias, Moritz$$b1$$eThesis advisor
000888741 7001_ $$0P:(DE-Juel1)156459$$aDahmen, David$$b2$$eThesis advisor
000888741 8564_ $$uhttps://juser.fz-juelich.de/record/888741/files/FischerKirsten_masterthesis.pdf$$yOpenAccess
000888741 909CO $$ooai:juser.fz-juelich.de:888741$$pdnbdelivery$$pdriver$$pVDB$$popen_access$$popenaire
000888741 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)180150$$aForschungszentrum Jülich$$b0$$kFZJ
000888741 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)144806$$aForschungszentrum Jülich$$b1$$kFZJ
000888741 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)156459$$aForschungszentrum Jülich$$b2$$kFZJ
000888741 9131_ $$0G:(DE-HGF)POF3-574$$1G:(DE-HGF)POF3-570$$2G:(DE-HGF)POF3-500$$3G:(DE-HGF)POF3$$4G:(DE-HGF)POF$$aDE-HGF$$bKey Technologies$$lDecoding the Human Brain$$vTheory, modelling and simulation$$x0
000888741 9141_ $$y2020
000888741 915__ $$0StatID:(DE-HGF)0510$$2StatID$$aOpenAccess
000888741 920__ $$lyes
000888741 9201_ $$0I:(DE-Juel1)INM-6-20090406$$kINM-6$$lComputational and Systems Neuroscience$$x0
000888741 9201_ $$0I:(DE-Juel1)IAS-6-20130828$$kIAS-6$$lTheoretical Neuroscience$$x1
000888741 9201_ $$0I:(DE-Juel1)INM-10-20170113$$kINM-10$$lJara-Institut Brain structure-function relationships$$x2
000888741 9801_ $$aFullTexts
000888741 980__ $$amaster
000888741 980__ $$aVDB
000888741 980__ $$aUNRESTRICTED
000888741 980__ $$aI:(DE-Juel1)INM-6-20090406
000888741 980__ $$aI:(DE-Juel1)IAS-6-20130828
000888741 980__ $$aI:(DE-Juel1)INM-10-20170113
000888741 981__ $$aI:(DE-Juel1)IAS-6-20130828